A method to evaluate and to reduce the source- and receiver- consistent noise in a cross-well travel time data was proposed. These systematic noises, which can cause some serious effects on the result of a travel time tomography, can be considered as the source and receiver statics. The method evaluates the statics through a curve-fitting of the first arrival travel times in the common source and common receiver gathers. Feasibility study was conducted on a synthetic data which simulates the cross-well travel time tomography to detect a small scale tunnel in a uniform background medium. First arrival travel times at a given source and receiver points are computed by a raytracing method, and the source consistent- and receiver consistent noises are added to the record. In case of the added noise with rms amounting to 25% of the maximum expected anomalous travel time delays, it is confirmed that the method successfully extracted the noise at the 7th step of iteration.